Building the Model on the Estimation of Pinus densata′s Biomass in Shangri-La City Based on Forest Subcompartment and Remote Sensing Images

The TM images of Shangri-La City in 2006, forest resource inventory data in 2006 and field survey data were adopted as the data source in this study. Sampling points were created randomly. Then, datasets were built through extracting subcompartment′s mean values based on remote sensing indexes. 78 sampling points (60 training data and 18 validation data) were selected by eliminating the abnormal values, and 14 indexes were collected as the alternative variables through index optimization. Stand volume was converted to forest biomass by a given model. Finally, a stepwise regression model and a partial least squares regression model for estimating Pinus densata′s biomass were established. The results showed that the stepwise regression model′s accuracy (R=0.518, RMSE=34.265 t/hm², rRMSE=47.046%) was higher than the partial least squares regression model′s (R=0.514, RMSE=35.320 t/hm², rRMSE=48.494%). The findings can provide a reference for the modeling of biomass based on remote sensing, planning and protecting ecological environment at high altitude.